A preliminary transcranial magnetic stimulation study of cortical inhibition and excitability in high-functioning autism and Asperger disorder
Dr Peter G Enticott at Monash Alfred Psychiatry Research Centre, Level 1, Old Baker Building, The Alfred, Melbourne, Victoria, 3004, Australia. E-mail: firstname.lastname@example.org
Aim Controversy surrounds the distinction between high-functioning autism (HFA) and Asperger disorder, but motor abnormalities are associated features of both conditions. This study examined motor cortical inhibition and excitability in HFA and Asperger disorder using transcranial magnetic stimulation (TMS).
Method Participants were diagnosed by experienced clinicians strictly according to DSM-IV criteria. Participants with HFA (nine males, two females; mean age 16y 8mo, SD 4y 5mo) or Asperger disorder (11 males, three females; mean age 19y 1mo, SD 4y 2mo) and neurotypical participants (eight males, three females; mean age 19y 0mo, SD 3y 1mo) were administered a paired-pulse TMS paradigm intended to assess motor cortical inhibition and excitability. Responses to TMS were recorded by electromyography.
Results Cortical inhibition was significantly reduced in the HFA group compared with both the Asperger disorder (p<0.001) and neurotypical (p<0.001) groups, suggesting disruption of activity at gamma-aminobutyric acid A (GABAA) receptors. There was no group difference in cortical excitability.
Interpretation Cortical inhibition deficits may underlie motor dysfunction in autism, and perhaps even relate to specific clinical symptoms (e.g. repetitive behaviours). These findings provide novel evidence for a possible neurobiological dissociation between HFA and Asperger disorder based on GABAergic function.
List of Abbreviations
Paired-pulse transcranial magnetic stimulation
Transcranial magnetic stimulation
What this paper adds
- • In vivo evidence for possible GABAergic dysfunction in autism (using TMS).
- • Evidence for a neurobiological dissociation between autism and Asperger disorder based on GABAergic function.
- • Further understanding of the neuropathophysiology of motor impairments and repetitive behaviours in autism.
- • Indication that possible neurochemical abnormalities beyond the motor cortices might underlie social and language impairments in autism.
Both autism and Asperger disorder involve social, communicative, motor, and behavioural disturbances (including stereotyped and repetitive behaviours), but there is controversy surrounding their clinical distinction,1 coupled with a relatively limited neurobiological understanding of the disorders. These conditions are primarily differentiated on the basis of language and communication delay, which must be present in autism but, by definition, is absent in Asperger disorder.2 In addition, although intelligence in individuals with Asperger disorder is typically within the average range, the majority of individuals with autism have intellectual disability. However, clinical research has demonstrated that the psychosocial profile of typical intelligent individuals with autism (i.e. high-functioning autism [HFA]) is convergent with that of individuals with Asperger disorder.3 Such findings form the basis of the argument that autism and Asperger disorder are not clinically distinct disorders, but are clinical labels that can be used interchangeably. By contrast, others have argued that the initial significant language delays that characterize autism (but not Asperger disorder) place the two disorders on separate neurodevelopmental trajectories.4 The hypothesis that autism and Asperger disorder are clinically separate disorders with distinct neurobiological underpinnings has been supported by neuropsychological and neurophysiological research that shows dissociations in visual perception, language, attention, executive function, and motor function.5
The neural circuitry involved in motor control is reasonably well defined; accordingly, studies of motor dysfunction in particular can further inform the debate concerning a possible neurobiological dissociation between autism and Asperger disorder. Although motor impairments are common to both autism and Asperger disorder, there is evidence to suggest that these impairments are neither behaviourally nor neurophysiologically equivalent. For example, analysis of gait function indicates that individuals with autism, compared with those with Asperger disorder and healthy individuals, show an increased variability in stride length, whereas those with Asperger disorder (but not autism) display significantly more head and trunk posturing abnormalities than comparison individuals.6 Movement dysfunction differences between autism and Asperger disorder have also been described in early childhood.7 Other studies, however, have found no difference between autism and Asperger disorder on standardized motor assessments8 and tests of coordination.9 Thus, although there may be behavioural differences in motor function, this area is controversial.
In relation to biology, recent brain imaging and neurophysiological studies of autism suggest dysfunction in both the basal ganglia/frontostriatal and cerebellar regions, but it is less clear whether this also extends to Asperger disorder. For example, Rinehart et al.10 and Enticott et al.11 report electroencephalographic (EEG) abnormalities over the supplementary motor area (similar to that found in Parkinson disease) in HFA but not in Asperger disorder, indicating a possible frontostriatal dissociation between the two disorders. Further, Muller et al.,12 in their functional magnetic resonance imaging (fMRI) study of autism, found decreased activation of motor-related neural circuitry (e.g. thalamus, basal ganglia, supplementary motor area), whereas subsequent fMRI research also found cerebellar abnormalities in a simple motor task in autism.13 Motor deficiencies in autism may, therefore, be underpinned by insufficient control of the basal ganglia/frontostriatal motor loop, perhaps attributable to reduced cortical inhibition arising from dysfunction in gamma-aminobutyric acid (GABA)-mediated processes. Indeed, there is some evidence for GABAergic abnormalities in autism.14,15 It is unclear, however, whether in vivo GABA impairments exist in autism and, if so, whether these impairments also extend to Asperger disorder (which is often considered a milder form of autistic disorder).
Aside from EEG investigations, and despite some neurobehavioural support for a frontostriatal dissociation, to date there have been no direct studies of motor cortical function (including GABAergic cortical inhibition) in autism and Asperger disorder. Transcranial magnetic stimulation (TMS) offers a unique tool with which to investigate motor cortical function in autism and Asperger disorder. To our knowledge, only one group has conducted a TMS study of the motor cortex in autism; Theoret et al.16 compared motor cortical function in 10 adults with ‘autism spectrum disorder’ (autism and Asperger disorder combined) and 10 comparison individuals, but found no between-group differences in cortical inhibition. HFA and Asperger disorder were not compared.
Using a well-established and validated paired-pulse TMS (ppTMS) paradigm, the aim of this study was to investigate cortical inhibition and excitability in individuals with HFA and Asperger disorder. By varying the time interval between pulses, ppTMS allows investigation of inhibitory (GABAergic)17 and excitatory (glutamatergic) cortical processes.18 Based on the body of experimental research indicating a dissociation in the nature of basal ganglia/frontostriatal dysfunction in autism and Asperger disorder, it was predicted that cortical inhibition would be more reduced in autism and relatively spared in Asperger disorder.
Participants included 11 individuals with HFA, 14 with Asperger disorder, and 11 neurotypical individuals (see Table I). Clinical participants were recruited via letters sent to parents of patients of one of the authors, while neurotypical participants were recruited via advertisements in the local community. All participants were diagnosed by experienced clinicians strictly according to DSM-IV criteria (autistic disorder [without intellectual disability] or Asperger disorder), while history of language development (delayed in HFA but not Asperger disorder) was also confirmed with a parent at the time of the TMS experiment. Participants were excluded if they had any additional psychiatric or neurological conditions. Seven participants with HFA were receiving low-dosage medication (one clomipramine 12.5mg, one risperidone 0.5mg, one quetiapine 100mg, one venlafaxine 150mg, two fluoxetine 20mg, and one valproate 600mg), and two participants with Asperger disorder were receiving medication (both fluoxetine 40mg). This study was approved by the ethics committees of The Alfred, Monash University, and Southern Health. Participants aged 18 years and over provided written informed consent. A parent or guardian provided written informed consent for participants aged below 18 years.
Table I. Demographic variables for all groups
|Age, mean (SD)||16y 8mo (4y 4mo)||19y 1mo (4y 2mo)||19y (3y 1mo)||0.289 (ANOVA)|
|Sex (M:F)||9:2||11:3||8:3||0.873 (χ2)|
TMS was administered to the motor cortex via a hand-held, 70mm figure-of-eight coil positioned over the scalp. Single-pulse TMS was administered using a Magstim 200 stimulator (Magstim Company Ltd., Carmarthenshire, Wales, UK); ppTMS was administered using two Magstim 200 stimulators linked with a bistim device. The coil was held tangential to the scalp, with the handle angled backwards and 45 degrees away from the midline. Electromyographic activity was recorded from the abductor pollicis brevis muscle, contralateral to the hemisphere receiving TMS, using self-adhesive electrodes.
Participants’ resting motor threshold and active motor threshold were determined for each hemisphere. Resting motor threshold was defined as the minimum stimulation intensity required to evoke a peak-to-peak motor-evoked potential (MEP) of more than 50μV in at least three out of five consecutive trials, whereas active motor threshold was defined as the minimum stimulation intensity required to produce an MEP of 100μV in at least one out of five trials during voluntary abductor pollicis brevis muscle contraction.
ppTMS involved the administration of a subthreshold (90% of active motor threshold) conditioning stimulus followed by a suprathreshold (115% resting motor threshold) test stimulus. The ppTMS paradigm was administered to each hemisphere, and involved a randomized sequence of 60 trials, comprising 20 of each of the following three conditions: single-pulse TMS, ppTMS with a 2ms interstimulus interval, and ppTMS with a 15ms interstimulus interval. There was a 5 second interval between each trial. This ppTMS protocol is widely used, and its effects are well established.19 Among healthy child and adult populations, a 2ms interval between the conditioning and test pulses produces an inhibitory effect (i.e. reduced MEP amplitude), reflecting GABAA function.17 By contrast, a 15ms interval results in a facilitatory effect (i.e. enhanced MEP amplitude), reflecting glutamatergic function.18 MEPs recorded for each ppTMS condition (2 and 15ms) were compared with MEPs recorded during single-pulse TMS.
Mixed-model analyses of variance (ANOVA; i.e. combined within- and between-individual data) were used to investigate differences in motor thresholds (resting and active) and MEP amplitudes. For motor thresholds, this involved two 2 (‘laterality’: left hemisphere vs right hemisphere) × 3 (‘group’: HFA vs Asperger disorder vs neurotypical) mixed-model ANOVAs (i.e. one for resting motor threshold, one for active motor threshold). For MEP amplitudes, there was an additional within-individual factor (i.e. TMS condition); thus, this involved a 2 (laterality: left hemisphere vs right hemisphere) × 3 (‘TMS’: single pulse vs 2ms ppTMS vs 15ms ppTMS) × 3 (group: HFA vs Asperger disorder vs neurotypical) mixed-model ANOVA. Least significant (LSD) post-hoc analyses were used to examine main effects. In addition, the response to ppTMS expressed as a percentage of the response to single-pulse TMS was calculated for each ppTMS condition (2 and 15ms), and compared between groups using a mixed-model ANOVA (with laterality as the within-individual factor, and LSD post-hoc testing where appropriate). For short-interval cortical inhibition (2ms ppTMS), a value below 100% is expected, whereas for cortical facilitation (15ms ppTMS) a value above 100% is expected. Data were analysed using SPSS version 16.0.1 (SPSS Inc., Chicago, IL, USA). Data were inspected to ensure adherence to the assumptions of ANOVA; violations of homogeneity of variance were noted for some analyses, and corrected values are reported. An alpha-level of 0.05 was adopted for all analyses.
Resting and active motor thresholds are presented in Table II. For resting motor thresholds, the main effect of group approached significance (F2,33=3.21, p=0.053). Post-hoc analyses indicated that the HFA group displayed a significantly greater threshold than the neurotypical group (p=0.018). There was no effect of laterality (F1,33=0.01; p=0.907), nor was there an interaction effect of laterality–group (F2,33=2.15; p=0.133). For active motor threshold, there was no effect of group (F2,33=2.44; p=0.103), laterality (F1,33=0.16; p=0.688), or laterality–group (F2,33=1.76; p=0.187).
Table II. Mean results for transcranial magnetic stimulation variables for all groups
|Resting motor threshold, %|
| Left hemisphere||50.9 (9.0)||49.6 (12.2)||42.7 (4.4)||0.104|
| Right hemisphere||53.1 (10.5)||47.6 (10.8)||42.2 (4.1)||0.030|
|Active motor threshold, %|
| Left hemisphere||42.4 (7.9)||41.2 (12.4)||35.1 (5.2)||0.158|
| Right hemisphere||43.7 (9.2)||39.2 (10.7)||34.8 (3.6)||0.068|
|Motor-evoked potentials – left hemisphere, mV|
| Single pulse||0.5 (0.3)||0.8 (0.7)||0.6 (0.4)||0.284|
| 2ms paired pulse||0.5 (0.4)||0.2 (0.2)||0.3 (0.2)||0.158|
| 15ms paired pulse||0.8 (0.4)||1.1 (0.8)||0.9 (0.6)||0.510|
|Motor-evoked potentials – right hemisphere, mV|
| Single pulse||0.5 (0.3)||0.8 (0.5)||0.7 (0.5)||0.324|
| 2ms paired pulse||0.4 (0.2)||0.3 (0.2)||0.3 (0.1)||0.218|
| 15ms paired pulse||0.8 (0.7)||1.2 (0.8)||1.0 (0.6)||0.498|
|Short-interval cortical inhibition, %|
| Left hemisphere||101 (51)||39 (22)||46 (28)||0.000|
| Right hemisphere||90 (41)||43 (34)||41 (23)||0.002|
|Cortical facilitation, %|
| Left hemisphere||212 (149)||180 (104)||159 (91)||0.561|
| Right hemisphere||185 (87)||162 (63)||150 (48)||0.457|
Cortical inhibition and excitability
MEP amplitudes are presented in Table II. TMS analyses failed Mauchley’s test of sphericity, and a Greenhouse–Geisser correction was used. There was an effect of TMS (F2,53=59.89; p<0.001) and a significant TMS–group interaction (F3,53=3.44; p=0.021). In relation to cortical inhibition, post-hoc analyses revealed no significant difference between single-pulse TMS and 2ms ppTMS among participants with HFA (p=0.311). By contrast, there was a significant difference between single-pulse TMS and 2ms ppTMS in both the Asperger disorder (p=0.001) and neurotypical (p=0.001) groups. In relation to cortical excitability (i.e. facilitation), there was a significant difference between single-pulse TMS and 15ms ppTMS in the HFA (p=0.014), Asperger disorder (p<0.001), and neurotypical (p=0.004) groups.
The cortical inhibition effects described above are further demonstrated when inspecting 2ms ppTMS MEP amplitudes expressed as a percentage of single-pulse TMS MEP amplitudes (per cent of reflecting short-interval cortical inhibition; see Table II). Mixed-model ANOVA revealed an effect of group (F2,33=15.28; p<0.001). Post-hoc analyses indicated that the per cent of short-interval cortical inhibition was significantly greater (i.e. less inhibition) in the HFA group than in either the Asperger disorder (p<0.001) or the neurotypical (p<0.001) group. There was no significant difference between the Asperger disorder group and the neurotypical group (p=0.788). There was no effect of laterality (F1,33=0.33; p=0.568) and no laterality–group interaction (F2,33=0.40; p=0.677). In relation to cortical facilitation, in which 15ms ppTMS amplitudes are expressed as a percentage of single-pulse TMS MEP amplitudes (reflecting cortical facilitation %; see Table II), there was no effect of group (F1,33=0.81; p=0.453), or laterality, (F1,33=1.26; p=0.270) and no group–laterality interaction (F2,33=0.10; p=0.903).
Motor dysfunction is apparent in both HFA and Asperger disorder, yet we have relatively little understanding of the motor cortical aspects of these disorders, and much controversy surrounds their separation in DSM-IV Text Revision (DSM-IV-TR).2 Reduced cortical inhibition during ppTMS in autism strongly suggests dysfunction of GABAergic circuits (specifically, impaired activity at GABAA receptors17) and supports the hypothesis that this disorder is associated with reduced control of the basal ganglia/frontostriatal motor loop.20 This is consistent with DSM-IV-TR accounts of motor dysfunction in autism (including stereotyped body movements, which seem to imply a deficit in inhibitory control), and empirical studies of the control of motor activity (including gait variability6) and motor-related brain activity.10–12 By contrast, cortical inhibition appears to be intact in Asperger disorder, suggesting that motor abnormalities (including clumsiness, reduced coordination, and impaired gait) may not be attributable to reduced motor cortical control. Cortical facilitation, largely reflecting glutamatergic processes,18 appears largely undifferentiated between HFA and Asperger disorder (and generally similar to neurotypical individuals, with the exception of right hemisphere resting motor threshold), despite evidence for cortical glutamatergic dysfunction in autism.21.
These findings represent perhaps the first direct functional brain evidence for a neurobiological dissociation between autism and Asperger disorder, and one presumably based on GABAergic function. Asperger disorder is frequently considered a ‘mild’ form of autism because of similarities in clinical presentation, but these findings instead provide support for the current clinically based distinction made in DSM-IV-TR.2 Based on these findings, it is conceivable that Theoret et al.16 did not detect differences in cortical inhibition between adults with autism spectrum disorder and comparison individuals because they were combining a group with normal (i.e. Asperger disorder) and abnormal (i.e. autism) cortical inhibition, which when combined would result in a ‘normal’ pattern of findings.
Although providing a more specific pathophysiological account of motor dysfunction, these results are also consistent with genetic and neurochemical research suggesting impaired GABAergic function in autism.14,15 Widespread abnormalities in GABAA receptor activity was recently demonstrated in post-mortem brain tissue samples from individuals with autism, with affected regions including parietal and frontal cortices and the cerebellum.22 These findings are also consistent with neural connectivity studies suggesting a reduction of inhibitory cells in autism,23,24 which operate as ‘GABA-gated pacemakers for neocortical oscillatory activity’ (Wilson et al.24 p 195).
Reduced cortical inhibition in the motor cortex may underlie some of the disorder-specific movement abnormalities seen in autism (e.g. an inability to anticipate when a movement is required25), but such cortical inhibitory deficits could also contribute to the specific executive function deficits that have been shown for autism (e.g. lower-level repetitive behavioural patterns), but not Asperger disorder. For example, repetitive behaviours in autism may result from impaired inhibition,26 whereas such behaviour in Asperger disorder could result from a failure to ‘spontaneously generate novel behaviour without prompting’27 (p 843) (reflecting higher-level repetitive behaviours, including circumscribed interests), although this hypothesis requires further investigation (see Rinehart et al.5 for a review).
This study is limited by a relatively small sample size, and its results should, therefore, be considered preliminary. A failure to include clinical measures to examine in relation to cortical inhibition is also a limitation and would have been of additional value. It might also be argued that standardized clinical measures should also have been included to differentiate HFA from Asperger disorder; however, we took the utmost care in reaching a diagnosis strictly according to DSM-IV criteria. Several HFA participants were receiving low-dose medication, but evidence suggests that the effect of these medications on GABAergic activity, if any, would be positive (i.e. increasing GABA-mediated cortical inhibition28). Theoretically, therefore, these medications limit our ability to detect GABAergic abnormalities in autism. Future investigations in this area should broaden the scope of TMS paradigms to include 100ms interstimulus interval ppTMS and a cortical silent period, both thought to involve GABAB function. This is particularly crucial given recent evidence of post-mortem GABAB abnormalities in autism.10 Larger sample sizes will also allow further exploration of cortical excitability and associations with clinical characteristics (e.g. repetitive behaviours, motor dysfunction), which should be measured in future studies. Nevertheless, the current findings represent a promising new direction for in vivo research elucidating the functional neuropathophysiology of autism and Asperger disorder.
The authors thank Dr Robin Laycock, Dr Sally Herring, and Mrs Hayley Kennedy for their assistance with data collection, and are extremely grateful to Ms Pamela Williams, Mr Dennis Freeman, and Wesley College Melbourne for assistance with participant recruitment. Funding was provided by Monash University (Small Grant Scheme) and the Cure Autism Now Foundation (Treatment-related Grant). Equipment funding was provided in part by Neurosciences Victoria (Clinical Neurobiology of Psychiatry Platform). PGE was supported by a NARSAD Young Investigator Award.